Mercurial > repos > genouest > helixer
changeset 2:e5ff71aed27b draft default tip
planemo upload for repository https://github.com/genouest/galaxy-tools/tree/master/tools/helixer commit 54445b041f46b8b3a8e25e90e84f034cdddd2ab1
| author | genouest |
|---|---|
| date | Tue, 09 Jul 2024 13:09:15 +0000 |
| parents | 0fefe93b5268 |
| children | |
| files | helixer.xml macros.xml test-data/test.h5 |
| diffstat | 2 files changed, 110 insertions(+), 69 deletions(-) [+] |
line wrap: on
line diff
--- a/helixer.xml Mon Sep 25 12:47:01 2023 +0000 +++ b/helixer.xml Tue Jul 09 13:09:15 2024 +0000 @@ -1,61 +1,76 @@ <?xml version="1.0"?> -<tool id="helixer" name="Helixer" version="@TOOL_VERSION@" profile="21.05"> +<tool id="helixer" name="Helixer" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="21.05"> <description>gene calling</description> <macros> <import>macros.xml</import> </macros> - + <xrefs> + <xref type="bio.tools">helixer</xref> + </xrefs> <requirements> <expand macro="requirements"/> </requirements> <command detect_errors="exit_code"><![CDATA[ - /usr/local/bin/fetch_helixer_models.py && Helixer.py --fasta-path '$input' --species '$species' - --lineage $lineage.lineages --gff-output-path '$output' - --temporary-dir ./ - --subsequence-length $lineage.subsequence_length - #if str($lineage.option_overlap.use_overlap) == "true": - --overlap-offset $lineage.option_overlap.overlap_offset - --overlap-core-length $lineage.option_overlap.overlap_core_length + #set default_subsequence_length = {"fungi": 21384, "land_plant": 106920, "invertebrate": 213840, "vertebrate": 213840} + #set subsequence_len = default_subsequence_length.get(str($lineages)) + #set default_overlap_offset = {"fungi": 10692, "land_plant": 53460, "invertebrate": 106920, "vertebrate": 106920} + #set overlap_off = default_overlap_offset.get(str($lineages)) + #set default_overlap_core_length = {"fungi": 16038, "land_plant": 80190, "invertebrate": 160380, "vertebrate": 160380} + #set overlap_core_len = default_overlap_core_length.get(str($lineages)) + + #if str($input_model) != "" and str($input_model) != "None": + --model-filepath '$input_model' + --subsequence-length '$subsequence_length' + #else: - --no-overlap + --lineage '$lineages' + #if str($subsequence_length) == "": + --subsequence-length '$subsequence_len' + #else: + --subsequence-length '$subsequence_length' + #end if + + #if str($option_overlap.use_overlap) == "true": + #if str($option_overlap.overlap_offset) == "": + --overlap-offset '$overlap_off' + #else: + --overlap-offset '$overlap_offset' + #end if + + #if str($option_overlap.overlap_core_length) == "": + --overlap-core-length '$overlap_core_len' + #else: + --overlap-core-length '$overlap_core_length' + #end if + #end if #end if + --batch-size $size --window-size $post_processing.window_size --min-coding-length $post_processing.min_coding_length --edge-threshold $post_processing.edge_threshold --peak-threshold $post_processing.peak_threshold + ]]></command> <inputs> <param argument="--fasta-path" name="input" type="data" format="fasta,fasta.gz" label="Genomic sequence"></param> - <conditional name="lineage"> - <param argument="--lineage" name="lineages" type="select" label="Available lineages" help="Choose the model to use for the annotation"> - <option value="land_plant">land plant</option> - <option value="vertebrate">vertebrate</option> - <option value="invertebrate">invertebrate</option> - <option value="fungi">fungi</option> - </param> - <when value="land_plant"> - <expand macro="subseq" length="106920" offset="53460" offsetlen="80190" /> - </when> - <when value="vertebrate"> - <expand macro="subseq" length="213840" offset="106920" offsetlen="160380" /> - </when> - <when value="invertebrate"> - <expand macro="subseq" length="213840" offset="106920" offsetlen="160380" /> - </when> - <when value="fungi"> - <expand macro="subseq" length="21384" offset="10692" offsetlen="16038" /> - </when> - </conditional> + <param argument="--model-filepath" optional="true" name="input_model" type="data" format="h5" label="Lineage model" help="Import your lineage model to replace the default lineage. Please enter a value for the --subsequence_length parameter."></param> + <param argument="--lineage" name="lineages" type="select" label="Available lineages" help="Choose the model to use for the annotation"> + <option value="land_plant">land plant</option> + <option value="vertebrate">vertebrate</option> + <option value="invertebrate">invertebrate</option> + <option value="fungi">fungi</option> + </param> + <param argument="--species" type="text" optional="true" label="Species name"> <sanitizer invalid_char=""> <valid initial="string.letters,string.digits"> @@ -71,6 +86,20 @@ <param argument="--peak-threshold" type="float" min="0" max="1" value="0.8" label="Peak threshold" help="This threshold specifies the minimum peak genic score required to accept the candidate region"/> <param argument="--min-coding-length" type="integer" min="0" value="100" label="Minimum coding length"/> </section> + + <param name="subsequence_length" type="text" label="Subsequence length: how much of the genome the Neural Network can see at once" help="If you do not want to keep the default value, please enter the new value. Default values are 21384 for fungi, 106920 for land plant, and 213840 for vertebrates and invertebrates"/> + + <conditional name="option_overlap"> + <param name="use_overlap" type="select" label="Enable overlapping step after predictions" help="This step combines predictions made on each subsequences, to improve quality near start and end of subsequences."> + <option value="true" selected="true">Yes</option> + <option value="false">No</option> + </param> + <when value="true"> + <param name="overlap_offset" type="text" label="Overlap offset: Smaller values may lead to better predictions but will take longer" help="If you do not want to keep the default value, please enter the new value. Default values are 10692 for fungi, 53460 for land plant, and 106920 for vertebrates and invertebrates."/> + <param name="overlap_core_length" type="text" label="Overlap core length: Predicted subsequences will be cut to this length to increase prediction quality. Smaller values may lead to better predictions but will take longer" help="If you do not want to keep the default value, please enter the new value. Default values are 16038 for fungi, 80190 for land plant, and 160380 for vertebrates and invertebrates."/> + </when> + <when value="false"/> + </conditional> </inputs> <outputs> @@ -79,41 +108,65 @@ </outputs> <tests> <test expect_num_outputs="1"> - <!-- Test for species and land_plant--> + <!-- Test for species and land_plant--> <param name="input" value="sequence.fasta"/> + <param name="lineages" value="land_plant"/> <param name="species" value="Arabidopsis"/> - <conditional name="lineage"> - <param name="lineages" value="land_plant"/> - </conditional> <param name="size" value="8"/> + <param name="subsequence_length" value=""/> + <section name="post_processing"> + <param name="window_size" value="100"/> + <param name="edge_threshold" value="0.1"/> + <param name="peak_threshold" value="0.8"/> + <param name="min_coding_length" value="100"/> + </section> + <conditional name="option_overlap"> + <param name="use_overlap" value="true"/> + <param name="overlap_offset" value=""/> + <param name="overlap_core_length" value=""/> + </conditional> <output name="output" value="ouput_species.gff3" ftype="gff3" compare="sim_size" delta="100"/> </test> + <test expect_num_outputs="1"> - <!-- Test for vertebrates--> + <!-- Test for vertebrates --> <param name="input" value="sequence.fasta"/> - <conditional name="lineage"> - <param name="lineages" value="vertebrate"/> - </conditional> + <param name="lineages" value="vertebrate"/> + <param name="size" value="8"/> <param name="size" value="8"/> + <param name="subsequence_length" value=""/> + <section name="post_processing"> + <param name="window_size" value="100"/> + <param name="edge_threshold" value="0.1"/> + <param name="peak_threshold" value="0.8"/> + <param name="min_coding_length" value="100"/> + </section> + <conditional name="option_overlap"> + <param name="use_overlap" value="true"/> + <param name="overlap_offset" value=""/> + <param name="overlap_core_length" value=""/> + </conditional> <output name="output" value="vertebrate.gff3" ftype="gff3" lines_diff="2"/> </test> - <test expect_num_outputs="1"> - <!-- Test for invertebrates--> + + <test expect_failure="true"> + <!-- Test for model-filepath --> <param name="input" value="sequence.fasta"/> - <conditional name="lineage"> - <param name="lineages" value="invertebrate"/> - </conditional> + <param name="input_model" value="test.h5"/> + <param name="size" value="8"/> <param name="size" value="8"/> - <output name="output" value="invertebrate.gff3" ftype="gff3" lines_diff="2"/> - </test> - <test expect_num_outputs="1"> - <!-- Test for fungi--> - <param name="input" value="sequence.fasta"/> - <conditional name="lineage"> - <param name="lineages" value="fungi"/> - </conditional> - <param name="size" value="8"/> - <output name="output" value="fungi.gff3" ftype="gff3" lines_diff="2"/> + <param name="subsequence_length" value="1000"/> + <section name="post_processing"> + <param name="window_size" value="100"/> + <param name="edge_threshold" value="0.1"/> + <param name="peak_threshold" value="0.8"/> + <param name="min_coding_length" value="100"/> + </section> + <conditional name="option_overlap"> + <param name="use_overlap" value="true"/> + <param name="overlap_offset" value=""/> + <param name="overlap_core_length" value=""/> + </conditional> </test> </tests>
--- a/macros.xml Mon Sep 25 12:47:01 2023 +0000 +++ b/macros.xml Tue Jul 09 13:09:15 2024 +0000 @@ -1,5 +1,7 @@ <macros> - <token name="@TOOL_VERSION@">0.3.2</token> + <token name="@TOOL_VERSION@">0.3.3</token> + <token name="@VERSION_SUFFIX@">1</token> + <xml name="citation"> <citations> @@ -11,20 +13,6 @@ <xml name="requirements"> <container type="docker">gglyptodon/helixer-docker:helixer_v@TOOL_VERSION@_cuda_11.8.0-cudnn8</container> </xml> - - <xml name="subseq" tokens="length,offset,offsetlen"> - <param argument="--subsequence-length" type="integer" min="0" max="213840" value="@LENGTH@" label="Subsequence length: how much of the genome the Neural Network can see at once" help="Should ideally be comfortably longer than the typical gene. For genomes with large genes (>20kpb) it is recommended to increase this parameter."></param> - <conditional name="option_overlap"> - <param name="use_overlap" type="select" label="Enable overlapping step after predictions" help="This step combines predictions made on each subsequences, to improve quality near start and end of subsequences"> - <option value="true" selected="true">Yes</option> - <option value="false">No</option> - </param> - <when value="true"> - <param argument="--overlap-offset" type="integer" min="0" value="@OFFSET@" label="Overlap offset" help="Smaller values may lead to better predictions but will take longer. The subsequence length should be evenly divisible by this value."/> - <param argument="--overlap-core-length" type="integer" min="0" value="@OFFSETLEN@" label="Overlap core length" help="Predicted subsequences will be cut to this length to increase prediction quality. Smaller values may lead to better predictions but will take longer. Has to be smaller than subsequence_length."/> - </when> - <when value="false"/> - </conditional> - </xml> + </macros>